Nvidia 副总裁 Deepu Talla:AI 将机器人训练环境生成速度提升 100 倍
Nvidia 机器人副总裁 Deepu Talla 在播客里讲了个实在数字:以前搭仿真训练环境要几个月,AI 生成后快了 100 倍。做机器人或具身智能开发的团队可以听听他们怎么把照片视频转成仿真世界,比自己搭环境省太多时间了。
Nvidia 机器人与边缘 AI 副总裁 Deepu Talla 在 The Information 的 AI Deep Dive 节目中表示,机器人行业最大的瓶颈之一是训练数据,而 AI 正在大幅加速这一环节。
其方法是让具备物理环境理解能力的 AI 智能体把照片和视频转换成可用于训练机器人的仿真世界,省去人工搭建环境的流程。Talla 给出的量化数据是:过去生成合成资产和环境需要数月时间,现在获得了 100 倍的速度提升。
这一变化直接影响机器人公司的开发节奏。合成数据生成是 Sim-to-Real 训练管线的关键环节,Nvidia 自身的 Cosmos 和 Isaac Sim 平台也围绕该方向布局,把数月级的环境构建压缩到天级,将降低机器人团队的训练成本和迭代周期。
AI could dramatically accelerate one of robotics' biggest bottlenecks: training data.
Nvidia Vice President of Robotics and Edge AI Deepu Talla explains how AI agents with an understanding of physical environments can turn photos and videos into simulated worlds for training robots, dramatically reducing the time needed to build those environments.
“What used to take months in generating synthetic assets and environments suddenly got 100x boost in speed up.”
🚀 Watch the full episode of AI Deep Dive on YouTube: https://t.co/YRpqJs4JWb
This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Nvidia, Deepu Talla, robots, synthetic data.
Source attribution: curated from The Information (@theinformation).